Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | Documentation for /proc/sys/kernel/* kernel version 2.2.10 |
2 | (c) 1998, 1999, Rik van Riel <riel@nl.linux.org> | |
760df93e | 3 | (c) 2009, Shen Feng<shen@cn.fujitsu.com> |
1da177e4 LT |
4 | |
5 | For general info and legal blurb, please look in README. | |
6 | ||
7 | ============================================================== | |
8 | ||
9 | This file contains documentation for the sysctl files in | |
10 | /proc/sys/kernel/ and is valid for Linux kernel version 2.2. | |
11 | ||
12 | The files in this directory can be used to tune and monitor | |
13 | miscellaneous and general things in the operation of the Linux | |
14 | kernel. Since some of the files _can_ be used to screw up your | |
15 | system, it is advisable to read both documentation and source | |
16 | before actually making adjustments. | |
17 | ||
18 | Currently, these files might (depending on your configuration) | |
19 | show up in /proc/sys/kernel: | |
807094c0 | 20 | |
1da177e4 | 21 | - acct |
807094c0 BP |
22 | - acpi_video_flags |
23 | - auto_msgmni | |
d75757ab PA |
24 | - bootloader_type [ X86 only ] |
25 | - bootloader_version [ X86 only ] | |
c114728a | 26 | - callhome [ S390 only ] |
73efc039 | 27 | - cap_last_cap |
1da177e4 | 28 | - core_pattern |
a293980c | 29 | - core_pipe_limit |
1da177e4 LT |
30 | - core_uses_pid |
31 | - ctrl-alt-del | |
eaf06b24 | 32 | - dmesg_restrict |
1da177e4 LT |
33 | - domainname |
34 | - hostname | |
35 | - hotplug | |
270750db AT |
36 | - hung_task_panic |
37 | - hung_task_check_count | |
38 | - hung_task_timeout_secs | |
39 | - hung_task_warnings | |
7984754b | 40 | - kexec_load_disabled |
455cd5ab | 41 | - kptr_restrict |
0741f4d2 | 42 | - kstack_depth_to_print [ X86 only ] |
1da177e4 | 43 | - l2cr [ PPC only ] |
ac76cff2 | 44 | - modprobe ==> Documentation/debugging-modules.txt |
3d43321b | 45 | - modules_disabled |
03f59566 | 46 | - msg_next_id [ sysv ipc ] |
1da177e4 LT |
47 | - msgmax |
48 | - msgmnb | |
49 | - msgmni | |
760df93e | 50 | - nmi_watchdog |
1da177e4 LT |
51 | - osrelease |
52 | - ostype | |
53 | - overflowgid | |
54 | - overflowuid | |
55 | - panic | |
807094c0 | 56 | - panic_on_oops |
55af7796 | 57 | - panic_on_stackoverflow |
9e3961a0 PB |
58 | - panic_on_unrecovered_nmi |
59 | - panic_on_warn | |
1da177e4 LT |
60 | - pid_max |
61 | - powersave-nap [ PPC only ] | |
62 | - printk | |
807094c0 BP |
63 | - printk_delay |
64 | - printk_ratelimit | |
65 | - printk_ratelimit_burst | |
1ec7fd50 | 66 | - randomize_va_space |
1da177e4 LT |
67 | - real-root-dev ==> Documentation/initrd.txt |
68 | - reboot-cmd [ SPARC only ] | |
69 | - rtsig-max | |
70 | - rtsig-nr | |
71 | - sem | |
03f59566 | 72 | - sem_next_id [ sysv ipc ] |
1da177e4 | 73 | - sg-big-buff [ generic SCSI device (sg) ] |
03f59566 | 74 | - shm_next_id [ sysv ipc ] |
b34a6b1d | 75 | - shm_rmid_forced |
1da177e4 LT |
76 | - shmall |
77 | - shmmax [ sysv ipc ] | |
78 | - shmmni | |
ed235875 | 79 | - softlockup_all_cpu_backtrace |
1da177e4 LT |
80 | - stop-a [ SPARC only ] |
81 | - sysrq ==> Documentation/sysrq.txt | |
f4aacea2 | 82 | - sysctl_writes_strict |
1da177e4 LT |
83 | - tainted |
84 | - threads-max | |
760df93e | 85 | - unknown_nmi_panic |
08825c90 | 86 | - watchdog_thresh |
1da177e4 LT |
87 | - version |
88 | ||
89 | ============================================================== | |
90 | ||
91 | acct: | |
92 | ||
93 | highwater lowwater frequency | |
94 | ||
95 | If BSD-style process accounting is enabled these values control | |
96 | its behaviour. If free space on filesystem where the log lives | |
97 | goes below <lowwater>% accounting suspends. If free space gets | |
98 | above <highwater>% accounting resumes. <Frequency> determines | |
99 | how often do we check the amount of free space (value is in | |
100 | seconds). Default: | |
101 | 4 2 30 | |
102 | That is, suspend accounting if there left <= 2% free; resume it | |
103 | if we got >=4%; consider information about amount of free space | |
104 | valid for 30 seconds. | |
105 | ||
807094c0 BP |
106 | ============================================================== |
107 | ||
108 | acpi_video_flags: | |
109 | ||
110 | flags | |
111 | ||
112 | See Doc*/kernel/power/video.txt, it allows mode of video boot to be | |
113 | set during run time. | |
114 | ||
115 | ============================================================== | |
116 | ||
117 | auto_msgmni: | |
118 | ||
119 | Enables/Disables automatic recomputing of msgmni upon memory add/remove | |
120 | or upon ipc namespace creation/removal (see the msgmni description | |
121 | above). Echoing "1" into this file enables msgmni automatic recomputing. | |
122 | Echoing "0" turns it off. auto_msgmni default value is 1. | |
123 | ||
124 | ||
1da177e4 LT |
125 | ============================================================== |
126 | ||
d75757ab PA |
127 | bootloader_type: |
128 | ||
129 | x86 bootloader identification | |
130 | ||
131 | This gives the bootloader type number as indicated by the bootloader, | |
132 | shifted left by 4, and OR'd with the low four bits of the bootloader | |
133 | version. The reason for this encoding is that this used to match the | |
134 | type_of_loader field in the kernel header; the encoding is kept for | |
135 | backwards compatibility. That is, if the full bootloader type number | |
136 | is 0x15 and the full version number is 0x234, this file will contain | |
137 | the value 340 = 0x154. | |
138 | ||
139 | See the type_of_loader and ext_loader_type fields in | |
140 | Documentation/x86/boot.txt for additional information. | |
141 | ||
142 | ============================================================== | |
143 | ||
144 | bootloader_version: | |
145 | ||
146 | x86 bootloader version | |
147 | ||
148 | The complete bootloader version number. In the example above, this | |
149 | file will contain the value 564 = 0x234. | |
150 | ||
151 | See the type_of_loader and ext_loader_ver fields in | |
152 | Documentation/x86/boot.txt for additional information. | |
153 | ||
154 | ============================================================== | |
155 | ||
c114728a HJP |
156 | callhome: |
157 | ||
158 | Controls the kernel's callhome behavior in case of a kernel panic. | |
159 | ||
160 | The s390 hardware allows an operating system to send a notification | |
161 | to a service organization (callhome) in case of an operating system panic. | |
162 | ||
163 | When the value in this file is 0 (which is the default behavior) | |
164 | nothing happens in case of a kernel panic. If this value is set to "1" | |
165 | the complete kernel oops message is send to the IBM customer service | |
166 | organization in case the mainframe the Linux operating system is running | |
167 | on has a service contract with IBM. | |
168 | ||
169 | ============================================================== | |
170 | ||
73efc039 DB |
171 | cap_last_cap |
172 | ||
173 | Highest valid capability of the running kernel. Exports | |
174 | CAP_LAST_CAP from the kernel. | |
175 | ||
176 | ============================================================== | |
177 | ||
1da177e4 LT |
178 | core_pattern: |
179 | ||
180 | core_pattern is used to specify a core dumpfile pattern name. | |
cd081041 | 181 | . max length 128 characters; default value is "core" |
1da177e4 LT |
182 | . core_pattern is used as a pattern template for the output filename; |
183 | certain string patterns (beginning with '%') are substituted with | |
184 | their actual values. | |
185 | . backward compatibility with core_uses_pid: | |
186 | If core_pattern does not include "%p" (default does not) | |
187 | and core_uses_pid is set, then .PID will be appended to | |
188 | the filename. | |
189 | . corename format specifiers: | |
190 | %<NUL> '%' is dropped | |
191 | %% output one '%' | |
192 | %p pid | |
65aafb1e | 193 | %P global pid (init PID namespace) |
b03023ec ON |
194 | %i tid |
195 | %I global tid (init PID namespace) | |
1da177e4 LT |
196 | %u uid |
197 | %g gid | |
12a2b4b2 ON |
198 | %d dump mode, matches PR_SET_DUMPABLE and |
199 | /proc/sys/fs/suid_dumpable | |
1da177e4 LT |
200 | %s signal number |
201 | %t UNIX time of dump | |
202 | %h hostname | |
57cc083a JS |
203 | %e executable filename (may be shortened) |
204 | %E executable path | |
1da177e4 | 205 | %<OTHER> both are dropped |
cd081041 MU |
206 | . If the first character of the pattern is a '|', the kernel will treat |
207 | the rest of the pattern as a command to run. The core dump will be | |
208 | written to the standard input of that program instead of to a file. | |
1da177e4 LT |
209 | |
210 | ============================================================== | |
211 | ||
a293980c NH |
212 | core_pipe_limit: |
213 | ||
807094c0 BP |
214 | This sysctl is only applicable when core_pattern is configured to pipe |
215 | core files to a user space helper (when the first character of | |
216 | core_pattern is a '|', see above). When collecting cores via a pipe | |
217 | to an application, it is occasionally useful for the collecting | |
218 | application to gather data about the crashing process from its | |
219 | /proc/pid directory. In order to do this safely, the kernel must wait | |
220 | for the collecting process to exit, so as not to remove the crashing | |
221 | processes proc files prematurely. This in turn creates the | |
222 | possibility that a misbehaving userspace collecting process can block | |
223 | the reaping of a crashed process simply by never exiting. This sysctl | |
224 | defends against that. It defines how many concurrent crashing | |
225 | processes may be piped to user space applications in parallel. If | |
226 | this value is exceeded, then those crashing processes above that value | |
227 | are noted via the kernel log and their cores are skipped. 0 is a | |
228 | special value, indicating that unlimited processes may be captured in | |
229 | parallel, but that no waiting will take place (i.e. the collecting | |
230 | process is not guaranteed access to /proc/<crashing pid>/). This | |
231 | value defaults to 0. | |
a293980c NH |
232 | |
233 | ============================================================== | |
234 | ||
1da177e4 LT |
235 | core_uses_pid: |
236 | ||
237 | The default coredump filename is "core". By setting | |
238 | core_uses_pid to 1, the coredump filename becomes core.PID. | |
239 | If core_pattern does not include "%p" (default does not) | |
240 | and core_uses_pid is set, then .PID will be appended to | |
241 | the filename. | |
242 | ||
243 | ============================================================== | |
244 | ||
245 | ctrl-alt-del: | |
246 | ||
247 | When the value in this file is 0, ctrl-alt-del is trapped and | |
248 | sent to the init(1) program to handle a graceful restart. | |
249 | When, however, the value is > 0, Linux's reaction to a Vulcan | |
250 | Nerve Pinch (tm) will be an immediate reboot, without even | |
251 | syncing its dirty buffers. | |
252 | ||
253 | Note: when a program (like dosemu) has the keyboard in 'raw' | |
254 | mode, the ctrl-alt-del is intercepted by the program before it | |
255 | ever reaches the kernel tty layer, and it's up to the program | |
256 | to decide what to do with it. | |
257 | ||
258 | ============================================================== | |
259 | ||
eaf06b24 DR |
260 | dmesg_restrict: |
261 | ||
807094c0 BP |
262 | This toggle indicates whether unprivileged users are prevented |
263 | from using dmesg(8) to view messages from the kernel's log buffer. | |
264 | When dmesg_restrict is set to (0) there are no restrictions. When | |
38ef4c2e | 265 | dmesg_restrict is set set to (1), users must have CAP_SYSLOG to use |
eaf06b24 DR |
266 | dmesg(8). |
267 | ||
807094c0 BP |
268 | The kernel config option CONFIG_SECURITY_DMESG_RESTRICT sets the |
269 | default value of dmesg_restrict. | |
eaf06b24 DR |
270 | |
271 | ============================================================== | |
272 | ||
1da177e4 LT |
273 | domainname & hostname: |
274 | ||
275 | These files can be used to set the NIS/YP domainname and the | |
276 | hostname of your box in exactly the same way as the commands | |
277 | domainname and hostname, i.e.: | |
278 | # echo "darkstar" > /proc/sys/kernel/hostname | |
279 | # echo "mydomain" > /proc/sys/kernel/domainname | |
280 | has the same effect as | |
281 | # hostname "darkstar" | |
282 | # domainname "mydomain" | |
283 | ||
284 | Note, however, that the classic darkstar.frop.org has the | |
285 | hostname "darkstar" and DNS (Internet Domain Name Server) | |
286 | domainname "frop.org", not to be confused with the NIS (Network | |
287 | Information Service) or YP (Yellow Pages) domainname. These two | |
288 | domain names are in general different. For a detailed discussion | |
289 | see the hostname(1) man page. | |
290 | ||
291 | ============================================================== | |
292 | ||
293 | hotplug: | |
294 | ||
295 | Path for the hotplug policy agent. | |
296 | Default value is "/sbin/hotplug". | |
297 | ||
298 | ============================================================== | |
299 | ||
270750db AT |
300 | hung_task_panic: |
301 | ||
302 | Controls the kernel's behavior when a hung task is detected. | |
303 | This file shows up if CONFIG_DETECT_HUNG_TASK is enabled. | |
304 | ||
305 | 0: continue operation. This is the default behavior. | |
306 | ||
307 | 1: panic immediately. | |
308 | ||
309 | ============================================================== | |
310 | ||
311 | hung_task_check_count: | |
312 | ||
313 | The upper bound on the number of tasks that are checked. | |
314 | This file shows up if CONFIG_DETECT_HUNG_TASK is enabled. | |
315 | ||
316 | ============================================================== | |
317 | ||
318 | hung_task_timeout_secs: | |
319 | ||
320 | Check interval. When a task in D state did not get scheduled | |
321 | for more than this value report a warning. | |
322 | This file shows up if CONFIG_DETECT_HUNG_TASK is enabled. | |
323 | ||
324 | 0: means infinite timeout - no checking done. | |
80df2847 | 325 | Possible values to set are in range {0..LONG_MAX/HZ}. |
270750db AT |
326 | |
327 | ============================================================== | |
328 | ||
70e0ac5f | 329 | hung_task_warnings: |
270750db AT |
330 | |
331 | The maximum number of warnings to report. During a check interval | |
70e0ac5f AT |
332 | if a hung task is detected, this value is decreased by 1. |
333 | When this value reaches 0, no more warnings will be reported. | |
270750db AT |
334 | This file shows up if CONFIG_DETECT_HUNG_TASK is enabled. |
335 | ||
336 | -1: report an infinite number of warnings. | |
337 | ||
338 | ============================================================== | |
339 | ||
7984754b KC |
340 | kexec_load_disabled: |
341 | ||
342 | A toggle indicating if the kexec_load syscall has been disabled. This | |
343 | value defaults to 0 (false: kexec_load enabled), but can be set to 1 | |
344 | (true: kexec_load disabled). Once true, kexec can no longer be used, and | |
345 | the toggle cannot be set back to false. This allows a kexec image to be | |
346 | loaded before disabling the syscall, allowing a system to set up (and | |
347 | later use) an image without it being altered. Generally used together | |
348 | with the "modules_disabled" sysctl. | |
349 | ||
350 | ============================================================== | |
351 | ||
455cd5ab DR |
352 | kptr_restrict: |
353 | ||
354 | This toggle indicates whether restrictions are placed on | |
312b4e22 RM |
355 | exposing kernel addresses via /proc and other interfaces. |
356 | ||
357 | When kptr_restrict is set to (0), the default, there are no restrictions. | |
358 | ||
359 | When kptr_restrict is set to (1), kernel pointers printed using the %pK | |
360 | format specifier will be replaced with 0's unless the user has CAP_SYSLOG | |
361 | and effective user and group ids are equal to the real ids. This is | |
362 | because %pK checks are done at read() time rather than open() time, so | |
363 | if permissions are elevated between the open() and the read() (e.g via | |
364 | a setuid binary) then %pK will not leak kernel pointers to unprivileged | |
365 | users. Note, this is a temporary solution only. The correct long-term | |
366 | solution is to do the permission checks at open() time. Consider removing | |
367 | world read permissions from files that use %pK, and using dmesg_restrict | |
368 | to protect against uses of %pK in dmesg(8) if leaking kernel pointer | |
369 | values to unprivileged users is a concern. | |
370 | ||
371 | When kptr_restrict is set to (2), kernel pointers printed using | |
372 | %pK will be replaced with 0's regardless of privileges. | |
455cd5ab DR |
373 | |
374 | ============================================================== | |
375 | ||
0741f4d2 CE |
376 | kstack_depth_to_print: (X86 only) |
377 | ||
378 | Controls the number of words to print when dumping the raw | |
379 | kernel stack. | |
380 | ||
381 | ============================================================== | |
382 | ||
807094c0 BP |
383 | l2cr: (PPC only) |
384 | ||
385 | This flag controls the L2 cache of G3 processor boards. If | |
386 | 0, the cache is disabled. Enabled if nonzero. | |
387 | ||
388 | ============================================================== | |
389 | ||
3d43321b KC |
390 | modules_disabled: |
391 | ||
392 | A toggle value indicating if modules are allowed to be loaded | |
393 | in an otherwise modular kernel. This toggle defaults to off | |
394 | (0), but can be set true (1). Once true, modules can be | |
395 | neither loaded nor unloaded, and the toggle cannot be set back | |
7984754b | 396 | to false. Generally used with the "kexec_load_disabled" toggle. |
3d43321b KC |
397 | |
398 | ============================================================== | |
399 | ||
03f59566 SK |
400 | msg_next_id, sem_next_id, and shm_next_id: |
401 | ||
402 | These three toggles allows to specify desired id for next allocated IPC | |
403 | object: message, semaphore or shared memory respectively. | |
404 | ||
405 | By default they are equal to -1, which means generic allocation logic. | |
406 | Possible values to set are in range {0..INT_MAX}. | |
407 | ||
408 | Notes: | |
409 | 1) kernel doesn't guarantee, that new object will have desired id. So, | |
410 | it's up to userspace, how to handle an object with "wrong" id. | |
411 | 2) Toggle with non-default value will be set back to -1 by kernel after | |
412 | successful IPC object allocation. | |
413 | ||
414 | ============================================================== | |
415 | ||
807094c0 BP |
416 | nmi_watchdog: |
417 | ||
418 | Enables/Disables the NMI watchdog on x86 systems. When the value is | |
419 | non-zero the NMI watchdog is enabled and will continuously test all | |
420 | online cpus to determine whether or not they are still functioning | |
421 | properly. Currently, passing "nmi_watchdog=" parameter at boot time is | |
422 | required for this function to work. | |
423 | ||
424 | If LAPIC NMI watchdog method is in use (nmi_watchdog=2 kernel | |
425 | parameter), the NMI watchdog shares registers with oprofile. By | |
426 | disabling the NMI watchdog, oprofile may have more registers to | |
427 | utilize. | |
428 | ||
429 | ============================================================== | |
430 | ||
10fc05d0 MG |
431 | numa_balancing |
432 | ||
433 | Enables/disables automatic page fault based NUMA memory | |
434 | balancing. Memory is moved automatically to nodes | |
435 | that access it often. | |
436 | ||
437 | Enables/disables automatic NUMA memory balancing. On NUMA machines, there | |
438 | is a performance penalty if remote memory is accessed by a CPU. When this | |
439 | feature is enabled the kernel samples what task thread is accessing memory | |
440 | by periodically unmapping pages and later trapping a page fault. At the | |
441 | time of the page fault, it is determined if the data being accessed should | |
442 | be migrated to a local memory node. | |
443 | ||
444 | The unmapping of pages and trapping faults incur additional overhead that | |
445 | ideally is offset by improved memory locality but there is no universal | |
446 | guarantee. If the target workload is already bound to NUMA nodes then this | |
447 | feature should be disabled. Otherwise, if the system overhead from the | |
448 | feature is too high then the rate the kernel samples for NUMA hinting | |
449 | faults may be controlled by the numa_balancing_scan_period_min_ms, | |
930aa174 | 450 | numa_balancing_scan_delay_ms, numa_balancing_scan_period_max_ms, |
52bf84aa | 451 | numa_balancing_scan_size_mb, and numa_balancing_settle_count sysctls. |
10fc05d0 MG |
452 | |
453 | ============================================================== | |
454 | ||
455 | numa_balancing_scan_period_min_ms, numa_balancing_scan_delay_ms, | |
930aa174 | 456 | numa_balancing_scan_period_max_ms, numa_balancing_scan_size_mb |
10fc05d0 MG |
457 | |
458 | Automatic NUMA balancing scans tasks address space and unmaps pages to | |
459 | detect if pages are properly placed or if the data should be migrated to a | |
460 | memory node local to where the task is running. Every "scan delay" the task | |
461 | scans the next "scan size" number of pages in its address space. When the | |
462 | end of the address space is reached the scanner restarts from the beginning. | |
463 | ||
464 | In combination, the "scan delay" and "scan size" determine the scan rate. | |
465 | When "scan delay" decreases, the scan rate increases. The scan delay and | |
466 | hence the scan rate of every task is adaptive and depends on historical | |
467 | behaviour. If pages are properly placed then the scan delay increases, | |
468 | otherwise the scan delay decreases. The "scan size" is not adaptive but | |
469 | the higher the "scan size", the higher the scan rate. | |
470 | ||
471 | Higher scan rates incur higher system overhead as page faults must be | |
472 | trapped and potentially data must be migrated. However, the higher the scan | |
473 | rate, the more quickly a tasks memory is migrated to a local node if the | |
474 | workload pattern changes and minimises performance impact due to remote | |
475 | memory accesses. These sysctls control the thresholds for scan delays and | |
476 | the number of pages scanned. | |
477 | ||
598f0ec0 MG |
478 | numa_balancing_scan_period_min_ms is the minimum time in milliseconds to |
479 | scan a tasks virtual memory. It effectively controls the maximum scanning | |
480 | rate for each task. | |
10fc05d0 MG |
481 | |
482 | numa_balancing_scan_delay_ms is the starting "scan delay" used for a task | |
483 | when it initially forks. | |
484 | ||
598f0ec0 MG |
485 | numa_balancing_scan_period_max_ms is the maximum time in milliseconds to |
486 | scan a tasks virtual memory. It effectively controls the minimum scanning | |
487 | rate for each task. | |
10fc05d0 MG |
488 | |
489 | numa_balancing_scan_size_mb is how many megabytes worth of pages are | |
490 | scanned for a given scan. | |
491 | ||
10fc05d0 MG |
492 | ============================================================== |
493 | ||
1da177e4 LT |
494 | osrelease, ostype & version: |
495 | ||
496 | # cat osrelease | |
497 | 2.1.88 | |
498 | # cat ostype | |
499 | Linux | |
500 | # cat version | |
501 | #5 Wed Feb 25 21:49:24 MET 1998 | |
502 | ||
503 | The files osrelease and ostype should be clear enough. Version | |
504 | needs a little more clarification however. The '#5' means that | |
505 | this is the fifth kernel built from this source base and the | |
506 | date behind it indicates the time the kernel was built. | |
507 | The only way to tune these values is to rebuild the kernel :-) | |
508 | ||
509 | ============================================================== | |
510 | ||
511 | overflowgid & overflowuid: | |
512 | ||
807094c0 BP |
513 | if your architecture did not always support 32-bit UIDs (i.e. arm, |
514 | i386, m68k, sh, and sparc32), a fixed UID and GID will be returned to | |
515 | applications that use the old 16-bit UID/GID system calls, if the | |
516 | actual UID or GID would exceed 65535. | |
1da177e4 LT |
517 | |
518 | These sysctls allow you to change the value of the fixed UID and GID. | |
519 | The default is 65534. | |
520 | ||
521 | ============================================================== | |
522 | ||
523 | panic: | |
524 | ||
807094c0 BP |
525 | The value in this file represents the number of seconds the kernel |
526 | waits before rebooting on a panic. When you use the software watchdog, | |
527 | the recommended setting is 60. | |
528 | ||
529 | ============================================================== | |
530 | ||
1da177e4 LT |
531 | panic_on_oops: |
532 | ||
533 | Controls the kernel's behaviour when an oops or BUG is encountered. | |
534 | ||
535 | 0: try to continue operation | |
536 | ||
a982ac06 | 537 | 1: panic immediately. If the `panic' sysctl is also non-zero then the |
8b23d04d | 538 | machine will be rebooted. |
1da177e4 LT |
539 | |
540 | ============================================================== | |
541 | ||
55af7796 MH |
542 | panic_on_stackoverflow: |
543 | ||
544 | Controls the kernel's behavior when detecting the overflows of | |
545 | kernel, IRQ and exception stacks except a user stack. | |
546 | This file shows up if CONFIG_DEBUG_STACKOVERFLOW is enabled. | |
547 | ||
548 | 0: try to continue operation. | |
549 | ||
550 | 1: panic immediately. | |
551 | ||
552 | ============================================================== | |
553 | ||
9e3961a0 PB |
554 | panic_on_unrecovered_nmi: |
555 | ||
556 | The default Linux behaviour on an NMI of either memory or unknown is | |
557 | to continue operation. For many environments such as scientific | |
558 | computing it is preferable that the box is taken out and the error | |
559 | dealt with than an uncorrected parity/ECC error get propagated. | |
560 | ||
561 | A small number of systems do generate NMI's for bizarre random reasons | |
562 | such as power management so the default is off. That sysctl works like | |
563 | the existing panic controls already in that directory. | |
564 | ||
565 | ============================================================== | |
566 | ||
567 | panic_on_warn: | |
568 | ||
569 | Calls panic() in the WARN() path when set to 1. This is useful to avoid | |
570 | a kernel rebuild when attempting to kdump at the location of a WARN(). | |
571 | ||
572 | 0: only WARN(), default behaviour. | |
573 | ||
574 | 1: call panic() after printing out WARN() location. | |
575 | ||
576 | ============================================================== | |
577 | ||
14c63f17 DH |
578 | perf_cpu_time_max_percent: |
579 | ||
580 | Hints to the kernel how much CPU time it should be allowed to | |
581 | use to handle perf sampling events. If the perf subsystem | |
582 | is informed that its samples are exceeding this limit, it | |
583 | will drop its sampling frequency to attempt to reduce its CPU | |
584 | usage. | |
585 | ||
586 | Some perf sampling happens in NMIs. If these samples | |
587 | unexpectedly take too long to execute, the NMIs can become | |
588 | stacked up next to each other so much that nothing else is | |
589 | allowed to execute. | |
590 | ||
591 | 0: disable the mechanism. Do not monitor or correct perf's | |
592 | sampling rate no matter how CPU time it takes. | |
593 | ||
594 | 1-100: attempt to throttle perf's sample rate to this | |
595 | percentage of CPU. Note: the kernel calculates an | |
596 | "expected" length of each sample event. 100 here means | |
597 | 100% of that expected length. Even if this is set to | |
598 | 100, you may still see sample throttling if this | |
599 | length is exceeded. Set to 0 if you truly do not care | |
600 | how much CPU is consumed. | |
601 | ||
602 | ============================================================== | |
603 | ||
55af7796 | 604 | |
1da177e4 LT |
605 | pid_max: |
606 | ||
beb7dd86 | 607 | PID allocation wrap value. When the kernel's next PID value |
1da177e4 LT |
608 | reaches this value, it wraps back to a minimum PID value. |
609 | PIDs of value pid_max or larger are not allocated. | |
610 | ||
611 | ============================================================== | |
612 | ||
b8f566b0 PE |
613 | ns_last_pid: |
614 | ||
615 | The last pid allocated in the current (the one task using this sysctl | |
616 | lives in) pid namespace. When selecting a pid for a next task on fork | |
617 | kernel tries to allocate a number starting from this one. | |
618 | ||
619 | ============================================================== | |
620 | ||
1da177e4 LT |
621 | powersave-nap: (PPC only) |
622 | ||
623 | If set, Linux-PPC will use the 'nap' mode of powersaving, | |
624 | otherwise the 'doze' mode will be used. | |
625 | ||
626 | ============================================================== | |
627 | ||
628 | printk: | |
629 | ||
630 | The four values in printk denote: console_loglevel, | |
631 | default_message_loglevel, minimum_console_loglevel and | |
632 | default_console_loglevel respectively. | |
633 | ||
634 | These values influence printk() behavior when printing or | |
635 | logging error messages. See 'man 2 syslog' for more info on | |
636 | the different loglevels. | |
637 | ||
638 | - console_loglevel: messages with a higher priority than | |
639 | this will be printed to the console | |
87889e15 | 640 | - default_message_loglevel: messages without an explicit priority |
1da177e4 LT |
641 | will be printed with this priority |
642 | - minimum_console_loglevel: minimum (highest) value to which | |
643 | console_loglevel can be set | |
644 | - default_console_loglevel: default value for console_loglevel | |
645 | ||
646 | ============================================================== | |
647 | ||
807094c0 BP |
648 | printk_delay: |
649 | ||
650 | Delay each printk message in printk_delay milliseconds | |
651 | ||
652 | Value from 0 - 10000 is allowed. | |
653 | ||
654 | ============================================================== | |
655 | ||
1da177e4 LT |
656 | printk_ratelimit: |
657 | ||
658 | Some warning messages are rate limited. printk_ratelimit specifies | |
659 | the minimum length of time between these messages (in jiffies), by | |
660 | default we allow one every 5 seconds. | |
661 | ||
662 | A value of 0 will disable rate limiting. | |
663 | ||
664 | ============================================================== | |
665 | ||
666 | printk_ratelimit_burst: | |
667 | ||
668 | While long term we enforce one message per printk_ratelimit | |
669 | seconds, we do allow a burst of messages to pass through. | |
670 | printk_ratelimit_burst specifies the number of messages we can | |
671 | send before ratelimiting kicks in. | |
672 | ||
673 | ============================================================== | |
674 | ||
807094c0 | 675 | randomize_va_space: |
1ec7fd50 JK |
676 | |
677 | This option can be used to select the type of process address | |
678 | space randomization that is used in the system, for architectures | |
679 | that support this feature. | |
680 | ||
b7f5ab6f HS |
681 | 0 - Turn the process address space randomization off. This is the |
682 | default for architectures that do not support this feature anyways, | |
683 | and kernels that are booted with the "norandmaps" parameter. | |
1ec7fd50 JK |
684 | |
685 | 1 - Make the addresses of mmap base, stack and VDSO page randomized. | |
686 | This, among other things, implies that shared libraries will be | |
b7f5ab6f HS |
687 | loaded to random addresses. Also for PIE-linked binaries, the |
688 | location of code start is randomized. This is the default if the | |
689 | CONFIG_COMPAT_BRK option is enabled. | |
1ec7fd50 | 690 | |
b7f5ab6f HS |
691 | 2 - Additionally enable heap randomization. This is the default if |
692 | CONFIG_COMPAT_BRK is disabled. | |
693 | ||
694 | There are a few legacy applications out there (such as some ancient | |
1ec7fd50 | 695 | versions of libc.so.5 from 1996) that assume that brk area starts |
b7f5ab6f HS |
696 | just after the end of the code+bss. These applications break when |
697 | start of the brk area is randomized. There are however no known | |
1ec7fd50 | 698 | non-legacy applications that would be broken this way, so for most |
b7f5ab6f HS |
699 | systems it is safe to choose full randomization. |
700 | ||
701 | Systems with ancient and/or broken binaries should be configured | |
702 | with CONFIG_COMPAT_BRK enabled, which excludes the heap from process | |
703 | address space randomization. | |
1ec7fd50 JK |
704 | |
705 | ============================================================== | |
706 | ||
1da177e4 LT |
707 | reboot-cmd: (Sparc only) |
708 | ||
709 | ??? This seems to be a way to give an argument to the Sparc | |
710 | ROM/Flash boot loader. Maybe to tell it what to do after | |
711 | rebooting. ??? | |
712 | ||
713 | ============================================================== | |
714 | ||
715 | rtsig-max & rtsig-nr: | |
716 | ||
717 | The file rtsig-max can be used to tune the maximum number | |
718 | of POSIX realtime (queued) signals that can be outstanding | |
719 | in the system. | |
720 | ||
721 | rtsig-nr shows the number of RT signals currently queued. | |
722 | ||
723 | ============================================================== | |
724 | ||
725 | sg-big-buff: | |
726 | ||
727 | This file shows the size of the generic SCSI (sg) buffer. | |
728 | You can't tune it just yet, but you could change it on | |
729 | compile time by editing include/scsi/sg.h and changing | |
730 | the value of SG_BIG_BUFF. | |
731 | ||
732 | There shouldn't be any reason to change this value. If | |
733 | you can come up with one, you probably know what you | |
734 | are doing anyway :) | |
735 | ||
736 | ============================================================== | |
737 | ||
358e419f CALP |
738 | shmall: |
739 | ||
740 | This parameter sets the total amount of shared memory pages that | |
741 | can be used system wide. Hence, SHMALL should always be at least | |
742 | ceil(shmmax/PAGE_SIZE). | |
743 | ||
744 | If you are not sure what the default PAGE_SIZE is on your Linux | |
745 | system, you can run the following command: | |
746 | ||
747 | # getconf PAGE_SIZE | |
748 | ||
749 | ============================================================== | |
750 | ||
807094c0 | 751 | shmmax: |
1da177e4 LT |
752 | |
753 | This value can be used to query and set the run time limit | |
754 | on the maximum shared memory segment size that can be created. | |
807094c0 | 755 | Shared memory segments up to 1Gb are now supported in the |
1da177e4 LT |
756 | kernel. This value defaults to SHMMAX. |
757 | ||
758 | ============================================================== | |
759 | ||
b34a6b1d VK |
760 | shm_rmid_forced: |
761 | ||
762 | Linux lets you set resource limits, including how much memory one | |
763 | process can consume, via setrlimit(2). Unfortunately, shared memory | |
764 | segments are allowed to exist without association with any process, and | |
765 | thus might not be counted against any resource limits. If enabled, | |
766 | shared memory segments are automatically destroyed when their attach | |
767 | count becomes zero after a detach or a process termination. It will | |
768 | also destroy segments that were created, but never attached to, on exit | |
769 | from the process. The only use left for IPC_RMID is to immediately | |
770 | destroy an unattached segment. Of course, this breaks the way things are | |
771 | defined, so some applications might stop working. Note that this | |
772 | feature will do you no good unless you also configure your resource | |
773 | limits (in particular, RLIMIT_AS and RLIMIT_NPROC). Most systems don't | |
774 | need this. | |
775 | ||
776 | Note that if you change this from 0 to 1, already created segments | |
777 | without users and with a dead originative process will be destroyed. | |
778 | ||
779 | ============================================================== | |
780 | ||
f4aacea2 KC |
781 | sysctl_writes_strict: |
782 | ||
783 | Control how file position affects the behavior of updating sysctl values | |
784 | via the /proc/sys interface: | |
785 | ||
786 | -1 - Legacy per-write sysctl value handling, with no printk warnings. | |
787 | Each write syscall must fully contain the sysctl value to be | |
788 | written, and multiple writes on the same sysctl file descriptor | |
789 | will rewrite the sysctl value, regardless of file position. | |
790 | 0 - (default) Same behavior as above, but warn about processes that | |
791 | perform writes to a sysctl file descriptor when the file position | |
792 | is not 0. | |
793 | 1 - Respect file position when writing sysctl strings. Multiple writes | |
794 | will append to the sysctl value buffer. Anything past the max length | |
795 | of the sysctl value buffer will be ignored. Writes to numeric sysctl | |
796 | entries must always be at file position 0 and the value must be | |
797 | fully contained in the buffer sent in the write syscall. | |
798 | ||
799 | ============================================================== | |
800 | ||
ed235875 AT |
801 | softlockup_all_cpu_backtrace: |
802 | ||
803 | This value controls the soft lockup detector thread's behavior | |
804 | when a soft lockup condition is detected as to whether or not | |
805 | to gather further debug information. If enabled, each cpu will | |
806 | be issued an NMI and instructed to capture stack trace. | |
807 | ||
808 | This feature is only applicable for architectures which support | |
809 | NMI. | |
810 | ||
811 | 0: do nothing. This is the default behavior. | |
812 | ||
813 | 1: on detection capture more debug information. | |
814 | ||
815 | ============================================================== | |
816 | ||
807094c0 | 817 | tainted: |
1da177e4 LT |
818 | |
819 | Non-zero if the kernel has been tainted. Numeric values, which | |
820 | can be ORed together: | |
821 | ||
bb20698d GKH |
822 | 1 - A module with a non-GPL license has been loaded, this |
823 | includes modules with no license. | |
824 | Set by modutils >= 2.4.9 and module-init-tools. | |
825 | 2 - A module was force loaded by insmod -f. | |
826 | Set by modutils >= 2.4.9 and module-init-tools. | |
827 | 4 - Unsafe SMP processors: SMP with CPUs not designed for SMP. | |
828 | 8 - A module was forcibly unloaded from the system by rmmod -f. | |
829 | 16 - A hardware machine check error occurred on the system. | |
830 | 32 - A bad page was discovered on the system. | |
831 | 64 - The user has asked that the system be marked "tainted". This | |
832 | could be because they are running software that directly modifies | |
833 | the hardware, or for other reasons. | |
834 | 128 - The system has died. | |
835 | 256 - The ACPI DSDT has been overridden with one supplied by the user | |
836 | instead of using the one provided by the hardware. | |
837 | 512 - A kernel warning has occurred. | |
838 | 1024 - A module from drivers/staging was loaded. | |
f5fe184b LF |
839 | 2048 - The system is working around a severe firmware bug. |
840 | 4096 - An out-of-tree module has been loaded. | |
66cc69e3 MD |
841 | 8192 - An unsigned module has been loaded in a kernel supporting module |
842 | signature. | |
69361eef | 843 | 16384 - A soft lockup has previously occurred on the system. |
1da177e4 | 844 | |
760df93e SF |
845 | ============================================================== |
846 | ||
760df93e SF |
847 | unknown_nmi_panic: |
848 | ||
807094c0 BP |
849 | The value in this file affects behavior of handling NMI. When the |
850 | value is non-zero, unknown NMI is trapped and then panic occurs. At | |
851 | that time, kernel debugging information is displayed on console. | |
760df93e | 852 | |
807094c0 BP |
853 | NMI switch that most IA32 servers have fires unknown NMI up, for |
854 | example. If a system hangs up, try pressing the NMI switch. | |
08825c90 LZ |
855 | |
856 | ============================================================== | |
857 | ||
858 | watchdog_thresh: | |
859 | ||
860 | This value can be used to control the frequency of hrtimer and NMI | |
861 | events and the soft and hard lockup thresholds. The default threshold | |
862 | is 10 seconds. | |
863 | ||
864 | The softlockup threshold is (2 * watchdog_thresh). Setting this | |
865 | tunable to zero will disable lockup detection altogether. | |
866 | ||
867 | ============================================================== |